Journal ofMaterials Chemistry A

PAPER

Publ

ishe

d on

13

June

201

4. D

ownl

oade

d by

Mon

ash

Uni

vers

ity o

n 25

/10/

2014

08:

48:5

4.

View Article OnlineView Journal | View Issue

aInstitut de Chimie et Procedes pour l'Energ

ECPM, UMR 7515 du CNRS-Universite d

Strasbourg Cedex 02, France. E-mail: cuongbInstitut de Physique et de Chimie des Materi

CNRS-Universite de Strasbourg, 23 rue du L

FrancecThe University of Da-Nang, University of Sc

Bang, Da-Nang, Viet NamdUnite de Catalyse et Chimie du Solide (UC

Lille-1, Batiment C3, Universite Lille 1, 596

† Electronic supplementary informa10.1039/c4ta01307g

Cite this: J. Mater. Chem. A, 2014, 2,11349

Received 18th March 2014Accepted 25th April 2014

DOI: 10.1039/c4ta01307g

www.rsc.org/MaterialsA

This journal is © The Royal Society of C

A few-layer graphene–graphene oxide compositecontaining nanodiamonds as metal-free catalysts†

Tung Tran Thanh,a Housseinou Ba,a Lai Truong-Phuoc,a Jean-Mario Nhut,a

Ovidiu Ersen,b Dominique Begin,a Izabela Janowska,a Dinh Lam Nguyen,c

Pascal Grangerd and Cuong Pham-Huu*a

We report a high yield exfoliation of few-layer-graphene (FLG) with up to 17% yield from expanded graphite,

under 5 h sonication time in water, using graphene oxide (GO) as a surfactant. The aqueous dispersion of

GO attached FLG (FLG–GO), with less than 5 layers, is used as a template for further decoration of

nanodiamonds (NDs). The hybrid materials were self-organized into 3D-laminated nanostructures, where

spherical NDs with a diameter of 4–8 nm are homogeneously distributed on the surface of the FLG–GO

complex (referred to as FLG–GO@NDs). It was found that GO plays a dual role, it (1) mediated exfoliation

of expanded graphite in aqueous solution resulting in a FLG–GO colloid system, and (2) incorporated ND

particles for the formation of composites. A high catalytic performance in the dehydrogenation of ethyl-

benzene on FLG–GO@ND metal-free catalyst is achieved; 35.1% of ethylbenzene conversion and 98.6%

styrene selectivity after a 50 h reaction test are observed which correspond to an activity of 896 mmolSTgcatalyst

�1 h�1, which is 1.7 and 5 times higher than those of the unsupported NDs and traditional

catalysts, respectively. The results demonstrate the potential of the FLG–GO@ND composite as a

promising catalyst for steam-free industrial dehydrogenation applications.

Introduction

Graphene and few-layer graphene (FLG), containing differentnumbers of graphene layers up to thirty, have attractedtremendous scientic interest due to their exceptional physicaland chemical properties which allow them to be applied inseveral potential applications.1–3 These materials can besynthesized by several methods, such as micromechanicalcleavage,4 epitaxial growth on silicon carbide or metalsurfaces,5–7 exfoliation of oxidized graphite and/or grapheneoxide (GO),8–11 mechanical ablation,12 and unzipping of multi-walled carbon nanotubes.13–16 Despite this large number ofefforts, mass production of graphene and FLG with a lownumber of graphene layers by a versatile and low-cost method isstill an active eld of research, especially solution-phase

ie, l'Environnement et la Sante (ICPEES),

e Strasbourg, 25 rue Becquerel, 67087

.pham-huu@unistra.fr

aux de Strasbourg (IPCMS), UMR 7504 du

œss, BP 43, 67034 Strasbourg Cedex 02,

ience and Technology, 54, Nguyen Luong

CS), UMR 8181 du CNRS-Universite de

55 – Villeneuve d'Ascq Cedex, France

tion (ESI) available. See DOI:

hemistry 2014

methods where graphene or FLG is directly obtained fromexfoliation of graphite precursors.17,18 This method, however,still has some drawbacks: (1) the synthesis is mostly carried outin organic solvents (e.g., MNP, DMF) which need to be carefullyrecycled,19,20 (2) the low exfoliation yield (1–3 wt.%) despite avery long sonication time (a few tens to hundred hours),21,22 and(3) in the case of exfoliation of graphite in aqueous solution, theuse of an amphiphilic surfactant may affect the quality andneeds further steps of treatments.23–25 In fact, direct exfoliationof hydrophobic graphite to graphene sheets in aqueous solutionwithout dispersing agents has been considered to be extremelydifficult or even unfeasible.26 The ideas based on the use of GOas a dispersion agent for stabilization of carbon materials, suchas CNTs and graphite, have been recently demonstrated in a fewstudies. For example, Kim and co-workers reported for the rsttime that GO, with its amphiphilic nature, can act as a surfac-tant to process insoluble materials of graphite in aqueoussolution.27,28 It has also been reported that GO can act as asurfactant to prevent the aggregation of CNTs during theprocess.29 Gogotsi and co-workers have performed an extensivereduction of GO attached on the expanded graphite (EG) surfacethrough microwave irradiation.30 However, the use of GO as aphysical agent for the synthesis of large scale single-to-few layergraphene, with a stable dispersion in a greener solvent, i.e.water, still remains a challenge to be tackled.

Synthetic NDs produced by detonation methods have anaverage size ranging from 4–10 nm, and their large surface-to-

J. Mater. Chem. A, 2014, 2, 11349–11357 | 11349

Journal of Materials Chemistry A Paper

Publ

ishe

d on

13

June

201

4. D

ownl

oade

d by

Mon

ash

Uni

vers

ity o

n 25

/10/

2014

08:

48:5

4.

View Article Online

volume ratio endows them with higher surface reactivity thanother forms of carbon. Together with a large surface area, NDsalso have high entropy, high structural defects and low toxicitywhich make them a promising candidate for several applica-tions in tribology, drug delivery, bio-imaging and tissue engi-neering just to cite a few of them.31 Hybrid materials consistingof graphene and nanodiamonds (NDs) could potentially displaynot only the unique properties of NDs and those of graphene,but also additional novel properties due to the synergisticeffects between them. Those materials are very interesting forfuture applications such as supercapacitors,32 photo-catalysts,33,34 and electrocatalysts.35 Recently, NDs have beenreported by Su and co-workers36 to be efficient metal-free cata-lysts in the steam-free dehydrogenation of ethyl-benzene intostyrene. The dehydrogenation mechanism can be described asfollows: activation of the alkane fragment of the ethylbenzeneby the ketone group to generate styrene along with the forma-tion of a carbon hydroxyl group as an intermediate followed bythe release of hydrogen and the consecutive generation of theformer C]O group. However, the NDs are used in powder formwhich is not well dispersed and leads to a considerable decreaseof the effective surface area for the catalytic process and alsoinduce a signicant pressure drop along the catalyst bed. Thusthe idea is to take advantage of the at surface of graphene-based materials to disperse the ND particles, which wouldincrease the effective contact surface area and the number ofactive sites, leading to an enhancement of their catalytic prop-erties. Wang et al.37 have reported that NDs are steadilydispersed on the reduced graphene oxide surface which leads toa signicant improvement of the specic capacitance of thecomposite owing to the high exposed surface area and shortdiffusion length of the NDs.

In this work, we have tried to address three main aspectsincluding: (1) direct exfoliation of expanded graphite inaqueous solutions (with high yield and short sonication time)using GO as the surfactant; (2) the use of GO attached few-layer-graphene (FLG–GO) aqueous dispersion as a template forfurther self-organized decoration of homogeneous nano-diamonds (NDs), and (3) such ND-decorated GO–FLG compos-ites (referred to as FLG–GO@NDs) will be used as metal-freecatalysts in the steam-free dehydrogenation (DH) of ethyl-benzene (EB) to styrene reaction which is one of the largestcatalytic processes nowadays. Herein, GO acts as an adsorbentlayer that concentrates ND nanoparticles on the surface of FLG–GO, meanwhile FLG plays the role of a reinforcement supportfor GO which is easily shrinkable during ND adsorption. Theresulting 3D-sandwich structure of the FLG–GO@ND compositeis tested as an advanced catalyst in the DH of EB to styrenereaction. The activities are compared with the ones obtainedfrom traditional catalysts. According to the results the FLG–GO@ND composite displays predominant performances alongwith a high stability as a function of time on stream comparedto the other catalysts. The present high-performance FLG–GO@ND composite with great advantages such as a simple andmass production method, and the absence of cytotoxicity couldbe foreseen as a potential metal-free catalyst in the steam freeindustrial synthesis of styrene.

11350 | J. Mater. Chem. A, 2014, 2, 11349–11357

ExperimentalExfoliation of expanded graphite in aqueous medium usinggraphene oxide as a surfactant

Graphene oxide was prepared by a modied Hummer's methodfrom ake graphite as described in the ESI.† A FLG–GO complexwas prepared by aqueous solution-processed exfoliation ofexpanded graphite in a GO suspension under ultrasonic treat-ment (see ESI†). Briey, a GO suspension with a concentrationof 1 mgmL�1 was obtained by dispersing GO powder in distilledwater with the aid of an ultrasonic bath for 30 min. Then thedifferent weight ratios of EG/GO, i.e. 1, 3, 5, were subjected tosonication with a tip sonicator of 30W for different time periodsranging from 1 to 5 h. Subsequently, the as-prepared suspen-sion was le to stand for 2 days for the precipitation of unstablegraphite aggregates. The supernatant was next extracted using apipette, and then centrifuged at 3000 rpm for 30 min to get thestable FLG–GO suspension which will be further processed.

Self-organized decoration of nanodiamonds on the FLG–GOsheets

The commercial nanodiamonds (NDs) with a diameter in therange of 4–10 nm in powder form were supplied by the HightechCo (Finland) and were used without any further purication.For preparation of the FLG–GO@ND composite, typically 200mg of pristine NDs were dispersed in 300 mL of deionized (DI)water followed by a sonication treatment for 30 min. It was thenwell mixed with 286 mL of the as-synthesized FLG–GO complexsuspension (0.35 mg mL�1) adding dropwise under stirringconditions. ND particles were steadily adsorbed in a homoge-neous manner on the surface of FLG–GO and the resultantcomposite settled down aer a few minutes. The adsorption ofthe NDs on the GO–FLG surface was evidenced by the change ofthe solution color from pale-yellow (ND suspension) to color-less. The mixture was then ltered and dried at 100 �C in anoven for 6 h to get a constant weight of the FLG–GO@NDcomposite.

Characterization techniques

UV-vis spectra were recorded using a spectrophotometerequipped with a Peltier PTP1 system (PerkinElmer Lambda 35)at room temperature. X-ray diffraction (XRD) data werecollected on a Bruker D8 Advance diffractrometer. X-rayphotoelectron spectroscopy (XPS) measurements were per-formed with an ESCA2000 (VG Microtech) spectrometer using amonochromatized aluminum Ka anode. Thermal gravimetricanalysis (TGA) was carried out on a TGA Q5000 instrument witha heating rate of 10 �C min�1 under an air atmosphere. Thespecic surface area was measured using the BET method fromthe nitrogen adsorption–desorption isotherm at 77 K (TriStarsorptometer). Prior to measurement, the sample was out-gassedat 250 �C during 12 hours in order to desorb the moisture and/or other impurities. The Raman spectra were recorded usingLabRAM ARAMIS Horiba Raman spectrometer equipment.Spectra were recorded over the range of 500–4000 cm�1 at alaser excitation wavelength of 532 nm. The sample was

This journal is © The Royal Society of Chemistry 2014

Fig. 1 (A) A digital photo of the different EG–GO mixtures in aqueousmedium after sonication for 3 h. (B) UV-vis spectra of the different EG–GO mixtures which show the degree of the EG exfoliation, (C and D)Raman spectra recorded on different samples according to the soni-cation time and the enlargement of the 2D peak.

Paper Journal of Materials Chemistry A

Publ

ishe

d on

13

June

201

4. D

ownl

oade

d by

Mon

ash

Uni

vers

ity o

n 25

/10/

2014

08:

48:5

4.

View Article Online

deposited on a glass substrate by spin-coating its suspensionand carefully dried before measurement. Scanning electronmicroscopy (SEM) images were recorded using a JEOL 2600Finstrument operating at an acceleration voltage of 15 kV and anemission current of 10 mA. The sample was covered with a thinlayer of sputtered gold in order to avoid charging problemsduring the analysis. Transmission electron microscopy (TEM)was conducted on a JEOL 2100F working at 200 kV acceleratedvoltage, equipped with a probe corrector for spherical aberra-tions, and a point-to-point resolution of 0.2 nm. The sample wasdispersed in ethanol with the aid of mild sonication for 5minutes and a drop of the suspension was deposited on a holeycarbon grid for measurement.

Catalytic activity measurements

The steam-free catalytic dehydrogenation of ethyl-benzene tostyrene was carried out in a tubular quartz reactor (74 cm lengthand 1.2 cm I.D.) under atmospheric pressure at 550 �C. Thecatalysts (300 mg) were deposited on a quartz wool plug and thereactant passed downward through the catalyst bed. Before thereaction, the catalyst was slowly heated up to the reactiontemperature and pretreated at 550 �C under helium for 2 h. Thereactant ow (2.8 vol% EB diluted in helium, with a total owrate of 30 mL min�1) was then fed to the reactor. The reactantsand the products at the exit from the reactor were analyzed on-line with a PERICHROM (PR 2100) gas chromatograph equip-ped with a ame ionization detector (FID). In order to avoid anypossible condensation of the reactant or products, all the tubelines were wrapped with a heating wire kept at 110 �C.

The ethyl-benzene conversion (XEB), styrene selectivity (SST)and yield (YST) were evaluated using the following equations:

XEB ¼ F0CEB;inlet � FCEB;outlet

F0CEB;inlet

� 100% (1)

SST ¼ CST;outlet

CST;outlet þ CTOL;outlet þ CBZ;outlet

� 100% (2)

YST ¼ XEB � SST (3)

where F and F0 are the inlet and outlet ow rates; CEB, CST, CTOL

and CBZ represent the concentration of ethyl benzene, styrene,toluene and benzene, respectively. The specic rate is expressedas the amount of styrene produced per gram of catalyst per hour(mmol g�1 h�1).

Results and discussion

The FLG was synthesized by mild sonication at 40 �C of aqueousmixtures containing EG and GO with different weight ratios, i.e.1 : 1, 3 : 1 and 5 : 1 (ESI, SI†). The sonication was done fordifferent durations and the products of these treatments werecollected for subsequent analysis. Fig. 1A shows the digitalphoto of the EG–GO mixtures aer 5 h of sonication, it high-lights the extremely high dispersion of the resultant materialswhen GO was added to the EG suspension: indeed no precipi-tation was observed aer several weeks of standing. On the

This journal is © The Royal Society of Chemistry 2014

other hand, the FLG formed from the EG suspension in water,without adding GO, and aer the same sonication treatment,steadily precipitates aer a few tens of minutes due to itshydrophobic nature and strong attraction forces. Similar resultshave also been observed by Hu et al.30 during the sonicationprocess with an EG–GO mixture with a high GO concentration.The high stability of the obtained FLG could be explained by anumber of physical methods that clearly demonstrated theproduction of single-to-few-layer graphene as shown in theschematic exfoliation process (schematic 1, ESI†).

The formation of a stable FLG–GO suspension as a functionof EG/GO was monitored by UV-vis spectroscopy as depicted inFig. 1B. The broad p–p* transition of the aromatic C]C peak ofGO centered at 230 nm�1 is gradually red shied several nm upto 238 nm when the EG/GO ratios increase from 0 to 5. Theabsorption intensity increases as well, and the highest intensityis observed with a mixture of EG/GO equal to 3; that shows thebest dispersion degree of EG/GO and the stabilization of exfo-liated FLG in the solution. When the EG/GO ratio is higher orlower than 3 (i.e., 5, 1), the absorption intensity tends todecrease, suggesting a lower concentration of chromophores inthe suspension, so a lower stabilization of the FLG sheets. Thedegree of exfoliation as a function of sonication time was alsoinvestigated for the EG/GO ratio equal to 3 as shown in Fig. S1A(ESI).† The amount of FLG in the FLG–GO complex increased upto 0.35mgmL�1 for 5 h sonication, and then exhibited a plateautrend for a longer sonication time (Fig. S1B†). The opticalabsorbance (A) divided by the length (l) as a function of theconcentration of the aqueous suspension of FLG–GO shows aLambert–Beer behavior, and a linear t of the concentrationvalues gives an average absorption coefficient at 660 nm of a ¼2244 L g�1 m�1 (Fig. S1C†), which is consistent with the valuereported previously for exfoliation of graphene in NMP.19

The XRD patterns of the as-synthesized materials as a func-tion of the EG–GO ratio are presented in Fig. S2.† The patterns

J. Mater. Chem. A, 2014, 2, 11349–11357 | 11351

Journal of Materials Chemistry A Paper

Publ

ishe

d on

13

June

201

4. D

ownl

oade

d by

Mon

ash

Uni

vers

ity o

n 25

/10/

2014

08:

48:5

4.

View Article Online

showed that the 2q peak of GO at 11� was gradually decreasedwhen the EG/GO ratio increased, while two peaks correspond-ing to (002) and (004) reections of graphite are increasedconsecutively. It means that the crystallinity of the FLG–GOcomplex increased as a function of EG/GO ratio, and when theEG/GO ratio reached 5, it would correspond to a large graphite-like structure in the products, which is consistent with the UV-vis measurements. The layer thickness and structural change ofthe FLG–GO lm as a function of sonication time, compared toits precursor materials (exfoliated EG and GO), were efficientlymonitored by Raman spectroscopy (Fig. 1C). While the Ramanspectrum shows a typical D-band at 1345 cm�1 (sp3 defects) anda G-band at 1592 cm�1 (tangential vibration of sp2 carbonatoms) for GO, those peaks appeared at 1350 and 1580 cm�1

respectively for FLG. A comparison of the intensity ratio of Dand G bands (ID/IG) as a function of sonication time for allsamples was also performed. The ID/IG decreased from 0.78 forpure GO to 0.24 for the FLG–GO complex aer 5 h sonicationconrming the better graphitic behavior of the sample. Inaddition, the G-band of the FLG–GO complex was red shied(8 cm�1) and blue shied (12 cm�1) as compared to that of FLGand GO, respectively, which conrms a physical interaction ofGO and the FLG sheets. The p-conjugated aromatic domainsexisting in the GO basal plane interact with the surface of FLGvia p–p interactions.28,30 Moreover, it is known that the 2D bandcorresponds to the specic number of stacked graphene lms.38

As compared to exfoliated EG, the 2D peak positions in FLG–GOshied considerably to a lower wavenumber by 25 cm�1, from2718 to 2693 cm�1, and the intensities of symmetric peaks arehigher than with FLG aer 3 h sonication, suggesting asuccessful exfoliation to less than 5 layers FLG in the FLG–GOcomplex.18,39 All of the above results have shown that GO sheets

Fig. 2 (A–C) Representative TEM micrographs of the GOmediated exfolGO and FLG–GO showing a strong reduction of the oxygen functional

11352 | J. Mater. Chem. A, 2014, 2, 11349–11357

signicantly promote the effective exfoliation of EG in aqueoussolution aer a relatively short time of sonication. The excellentwater solubility of GO made the FLG–GO complex to have amicelle system-like behavior.

Representative TEM micrographs of the as-synthesized FLGdecorated with GO sheets are presented in Fig. 2. They conrmthe high degree of exfoliation of the process as almost FLGcontaining less than ve graphene layers is observed among thesample (Fig. 2A). High-resolution TEM micrographs alsoevidence the non-covalent stacking between the FLG and GO aspresented in Fig. 2B and C, where a perfect graphitized struc-ture of the FLG is visible underneath the disordered structure ofthe GO. The nature and concentration of the oxygen speciespresent on the different samples were also analyzed by XPS andthe results are presented in Fig. 2D and E. The C1s XPS spec-trum of GO, with an oxygen content of �32.6 at.%, is presentedin Fig. 2D and can be deconvoluted into four peaks corre-sponding to carbon atoms in different functional groups: the C–C in aromatic rings, the oxygenated C in C–O, C]O, and O–C]O. However, the high percentage of oxygen in GO was steadilydecreased in the FLG–GO complex (Fig. 2E), with an oxygencontent of �8.3 at.%. In addition, the ratio of C/O obtainedfrom the C1s and O1s peaks increased from 2 (for GO) to 10 (forFLG–GO), respectively. Such a result could be attributed to theintimate mixing between the GO and the exfoliated FLG withlow oxygen content. It is expected that during the sonicationprocess, a part of the oxygen content could also be removedfrom the GO surface due to the energy input.

The yield of the exfoliation process as a function of thesonication duration is presented in Fig. 3. The FLG exfoliationyield was calculated based on the weight of initial and nalmaterials as dened in eqn (4):

iated EG with different magnifications. (D and E) C1s XPS spectra of thegroups along with the increase of the carbon species after mixing.

This journal is © The Royal Society of Chemistry 2014

Fig. 4 (A) A digital photo of the aqueous dispersion of pristine NDs(left), FLG–GO complex suspension (middle) and ND-adsorbed FLG–GO (right). (B) XRD spectra, (C) Raman spectrum and (D) C1s XPSspectrum of the FLG–GO@ND composite.

Paper Journal of Materials Chemistry A

Publ

ishe

d on

13

June

201

4. D

ownl

oade

d by

Mon

ash

Uni

vers

ity o

n 25

/10/

2014

08:

48:5

4.

View Article Online

exfð%Þ ¼ iniðEGþGOÞ � ðunexfEGþ setl:exfMLG@GOÞiniðEGþGOÞ

� 100% (4)

where ini(EG + GO) is the initial mass sum of EG and GO beforesonication, unexfEG is the un-exfoliated EG decanting aer 2days standing, and setl.exfMLG@GO is the GO adsorbed onmulti-layer-graphene that settled down aer centrifugation at3000 rpm. The exfoliation of EG that was calculated accordingto eqn (4) increases from 2.2, 5.3, 8.8, 12.3 to 16.8 wt% relativeto the initial raw materials (EG added GO) for the sonicationtime ranging from 1, 2, 3, 4 to 5 hours, respectively. The exfo-liation yield was also measured by UV-vis on the differentconcentrations of FLG–GO suspensions (Fig. 3B). The red line isa linear t in which the yield is relative to initial EG (if assumedthat no GO was decanted during the synthesis process), whilethe blue line is the yield relative to the total EG–GO mass. Theexfoliation yield increases with sonication time, and reaches thevalue of up to �25 wt% relative to the initial EG aer 5h soni-cation. The GO mass percentage in the FLG–GO mixture can beestimated to be 25 wt % aer 5 h sonication. These data areconsistent with TGA measurement as shown in Fig. S3,† inwhich the curve corresponding to the GO has the onsettemperature of 130 and 450 �C, which can be related to thedecomposition of different oxygen containing groups, while theonset temperatures of the GO–FLG complex are increased andwas found to be 360 and 560 �C. The amount of GO in the GO–FLG complex is estimated to be about 24% by mass. Accordingto the results one can state that the GO mediated EG exfoliationprocess is among the best method to obtain thin FLG with highyield and under mild treatment conditions.

From the reported results, we have shown that a thin FLG–GO complex was successfully prepared in aqueous suspensionmedium by the solution-phase method. The as-synthesizedFLG–GOmixture is used as a template for further self-organizeddecoration of FLG–GO with homogeneous nanodiamonds(NDs), for application as a metal-free catalyst in the steam-freedehydrogenation (DH) of ethyl benzene (EB) to styrene reaction.

Fig. 4A shows a digital photo of the aqueous dispersion ofpristine NDs aer 15 min tip sonication (Fig. 4A, le), and acolloidal suspension of the FLG–GO (3–1) complex (Fig. 4B,middle). When the ND dispersion (2 parts) is dropped into theFLG–GO suspension (1 part) under stirring conditions, NDparticles are steadily adsorbed onto the surface of the FLG–GO

Fig. 3 Exfoliation yield of EG calculated with eqn 4 (A), data calculatedfrom the maximum concentration relative to starting materials and thelinear fit (B).

This journal is © The Royal Society of Chemistry 2014

complex, which results in the precipitation of the FLG–GO@ND(1 : 2) product in water (Fig. 4C, right). The complete adsorptionof the NDs was evidenced by the colour change from pale-yellowto colourless of the medium upon precipitation. Herein, theaqueous dispersion of NDs has a positive charge with a zetapotential of +16 mV, whereas the FLG–GO suspension is nega-tively charged (�46 mV); thus, the interaction between themcan be considered as electrostatic attractions (a zeta potential ofthe re-dispersed FLG–GO@ND suspension shows a value of�28mV (Fig. S4A†)). Other possible interactions between FLG–GOand NDs may be hydrogen-bonding attractions because of thepresence of oxygen-containing groups including hydroxyl,carboxylic, lactones, ketones and ethers31,40 on the surface ofNDs which can strongly interact with the GO surface.

It should be noted that NDs are not adsorbed onto thesurface of FLG under similar conditions (Fig. S4B†). Themicrostructure of the FLG–GO@ND composite was investigatedby SEM (Fig. S5†), which shows that the FLG–GO powderconsists of thin FLG platelets covered by a slightly wrinkled GOlayer. The FLG–GO@NDs showed 3D laminated structureswhere the GO sheets were homogeneously decorated with NDparticles. The XRD pattern of FLG–GO@NDs is shown inFig. 4B, while the strong diffraction peak at 2q ¼ 26.5� isattributed to the (002) reection from the graphite, the otherpeaks correspond to the self-organized adsorption of NDs ontothe surface of the FLG–GO support. Raman spectroscopy at lexc¼ 532 nm was employed to investigate a change in the graphiticstructure aer adsorption of NDs, as shown in Fig. 4C. TheRaman spectrum of FLG–GO@NDs shows an increased inten-sity and a broadening of sp3 defects (D-band) as compared tothe FLG–GO complex since it overlapped with a peak of NDsitself (�1324 cm�1). The tangential vibration of sp2 carbonatoms in the hexagonal plane (G-band) is signicantlydecreased, and the ID/IG ratio increases more than 3 times from0.24 to 0.74. It reects that NDs cover the surface of FLG–GOand that the vibration of the graphitic domains is considerably

J. Mater. Chem. A, 2014, 2, 11349–11357 | 11353

Journal of Materials Chemistry A Paper

Publ

ishe

d on

13

June

201

4. D

ownl

oade

d by

Mon

ash

Uni

vers

ity o

n 25

/10/

2014

08:

48:5

4.

View Article Online

attenuated. In addition, the G-band shied by about 10 cm�1 toa higher wavenumber, which may be attributed to the func-tional groups on the surface of NDs such as the O–H bendingand C]O stretching.31 The composition of the FLG–GO@NDcomposite was analyzed by XPS and the results are presented inFig. 4D. The C1s XPS spectrum clearly indicates four differentmajor components (C sp2, C sp3, C–O and O]C–O), in whichthe non-oxygenated sp2 carbon of graphene is signicantlydecreased when compared to the FLG–GO complex, whereas thesp3 allotrope of carbon at 286.9 eV becomes a prominent peak.This peak corresponds to the ND incorporation.

The dispersion of the NDs on the surface of the FLG–GO hasbeen characterized by TEMmicroscopy (Fig. 5). It is clearly seenthat the FLG–GO sheets are homogeneously decorated byspherical ND particles. Herein, the GO acts as an adsorptionlayer for randomly distributed NDs on the surface of the FLG–GO complex, while FLG plays the role of an inactive support tostrengthen the FLG–GO complex, i.e. to hinder the GO layershrinkage during ND adsorption. The high anchorage strengthof the NDs on the GO surface could be attributed to the pres-ence of OH groups on the ND surface. Wang et al.37 havereported that those OH groups also actively participate in thereduction process of the GO aer further heating treatments.

From higher magnication TEM images, it can be seen thatthe ND nano-crystallites with an average diameter ranging from4 to 8 nm are isolated over FLG–GO sheets. A distribution of NDparticle sizes has been measured using the Image-Pro Plusprogram on TEM images and presented in Fig. S6.† This is animportant factor for a signicant improvement of catalystperformances of the FLG–GO@NDs which will be considered inthe following.

In order to evaluate the catalytic performance of the FLG–GOsupported ND catalyst in the dehydrogenation of ethyl-benzene,300 mg of the composite material were placed at the centre of aquartz reactor. The reaction was carried out at 550 �C underatmospheric pressure conditions similar to those reported by

Fig. 5 Representative TEM micrographs of the FLG–GO@NDcomposite with different magnifications. (A) General view showing thehomogeneous dispersion of the NDs on the composite surface. (B)Medium view evidencing the high affinity of the NDs to adsorb on theGO surface. (C and D) High-resolution views showing the micro-structure of the NDs and a detail view of the NDs' top surface.

11354 | J. Mater. Chem. A, 2014, 2, 11349–11357

Su et al.36 The reaction rate was expressed as the amount ofstyrene produced per gram of catalyst per hour. The catalystselectivity to styrene has also been investigated. Fig. 6A shows acomparison of the styrene formation rate (ST. rate) and styreneselectivity (ST. Sel.) on the FLG–GO@ND composite-basedcatalyst versus those obtained on unsupported NDs andcommercial iron-based catalysts (denoted as K–Fe). It wasobserved that the K–Fe catalyst exhibits extremely high DHactivity at the beginning of the test (853 mmolST gcatalyst

�1 h�1)followed by a sharp deactivation to be next stabilized aer 8 h ofreaction at around 180 mmolST gcatalyst

�1 h�1. The low DHactivity obtained on the iron-based catalyst could be attributedto the rapid saturation of the active sites by the carbonaceousresidue under steam-free reaction conditions. On the otherhand, both the unsupported NDs and FLG–GO@ND composite-based catalysts display a signicant improvement of the DHactivity, compared to the K–Fe catalyst, with a gradual decreasefrom 1064 to 402 mmolST gcatalyst

�1 h�1 and 1680 to 687 mmolSTgcatalyst

�1 h�1, respectively, aer the same duration of test, i.e. 50h of time on stream. However, in contrast to the commercialcatalyst, ND-based catalysts display a slow deactivation aer therst deactivation slope. Such observation indicates that siteblockage as a function of time on stream still occurs unlikely tothe industrial catalyst where a balance of site blockage by thedeposited carbonaceous residue was reached aer a few hoursof reaction. The signicant improvement of the DH activity perunit weight of the NDs obtained on the supported FLG–GOcatalyst could be attributed to the higher dispersion of the NDson the GO surface, as evidenced by SEM and TEM analysespresented above. The dehydrogenation rate as a function oftime on stream observed on both catalysts, NDs and FLG–GO@NDs, is similar which indicates that the nature of theactive site is unchanged and the DH activity improvement ismostly due to the higher active sites on the FLG–GO@NDsconsecutive to the dispersion of the NDs on the FLG–GOsurface. This provides a higher specic surface area and densityof active sites to perform the dehydrogenation reaction. TheBET value increased from 283 m2 g�1 for pristine NDs to 304 m2

g�1 for FLG–GO@NDs along with a signicant rise in meso-porous contribution (Fig. S7†). In the FLG–GO@ND catalyst theFLG with higher mechanical strength plays a role of an inactivesupport to prevent the clumping of the catalyst. To conrm ourdiscussion on the role of FLG in the composite, a GO@ND(without FLG) sample was tested under the same reactionconditions (Fig. S8†). The GO@ND catalyst exhibits an inter-mediate DH activity compared to the ND and FLG–GO@NDcatalysts. The low DH activity on the GO@ND catalyst might belinked to the fact that GO clumped during adsorption of NDs,without FLG underneath, which will be limited to the density ofactive sites and thus, leading to the lower DH performance.

The steady-state DH activity of the FLG–GO@ND catalyst wascompared also with those obtained on different carbon-basedand K–Fe catalysts (Fig. 6B). The results evidence the high DHactivity of the FLG–GO@ND catalyst compared to the othercarbon-based nanomaterial catalysts such as stacked GO, thickFLG, and CNTs. The high DH activity observed on the ND-basedcatalyst could be attributed to the high effective surface area of

This journal is © The Royal Society of Chemistry 2014

Fig. 6 (A) DH activity of FLG–GO@NDs in comparison with commercial, unsupported ND catalysts, (B) benchmarking of different carbon-basednanomaterials in the DH process, (C) long-term DH test at 550 �C of the FLG–GO@ND catalyst, and (D) regeneration test during cyclic air-exposure showing a high DH activity recovery of the GO–FLG@ND-based composite catalyst.

Paper Journal of Materials Chemistry A

Publ

ishe

d on

13

June

201

4. D

ownl

oade

d by

Mon

ash

Uni

vers

ity o

n 25

/10/

2014

08:

48:5

4.

View Article Online

the catalyst and also to the high stability of the oxygenatedfunctional groups on the ND surface compared to that exists onthe other carbon-based materials.

The stability of the FLG–GO@ND catalyst as a function oftime on stream is also evaluated by performing a long-term testup to 150 h (Fig. 6C). The DH activity was stabilized at around500 mmolST gcatalyst

�1 h�1, aer 150 h of test, while the selec-tivity towards styrene remains unchanged for the whole test.Such results indicate that the deactivation was mostly linkedwith the loss of active sites and not to a modication of theactive site nature.

The DH activity can be recovered by submitting the spentcatalyst to the regeneration process in air at 400 �C for 2 h(Fig. 6D). It is also worth noting that successive regenerationscontribute also to a better stability of the catalyst as a functionof time on stream, as evidenced in Fig. 6D. Similar behavior hasbeen also reported by Su and co-workers36 in a previous report.

In order to understand the cause of the deactivation and theregeneration mechanism, the XPS analysis was performed onthe catalyst at different stages, i.e. fresh, spent and regeneratedcatalyst, and the results are shown in Fig. S9.† As can be seenfrom the XPS data, the carbon-to-oxygen ratio on the catalyststeadily increases from 13.8 for the fresh catalyst to 18.6 for theone aer reaction. Such results indicate that carbonaceousspecies were deposited onto the catalyst surface during thecourse of the reaction which slowly covers the surface of theactive sites leading to catalyst deactivation. It is also worthnoting that during the reaction part of the oxygen speciespresents on the catalyst surface, especially the quinoic group

This journal is © The Royal Society of Chemistry 2014

which is expected to be the active site, could be also removed.The C/O ratio steadily decreases from 18.6 to 7.7 aer theregeneration process indicating that part of the carbonaceousresidues has been removed, leading to a higher active siteexposure. The change in the oxygen-containing functionalgroups, such as C–O and O–C]O, is well matched with the DHactivity behaviour observed on the catalyst performances.According to the results, the decrease of the oxygenated func-tional groups, especially the carboxylate group, is well ttedwith the decrease of the DH activity of the catalyst. The C]Ogroups slightly decrease from 11 to 8 at.% which corresponds tothe decrease of the DH activity from 1680 mmolST gcatalyst

�1 h�1

to 687 mmolST gcatalyst�1 h�1. Aer regeneration, the C]O

functional groups steadily increase from 8 to about 57 at.%while the DH activity is fully restored (Fig. 6D). Our results areconsistent with those previously reported by Su and co-workersin which the reduced DH activity agrees well with theconsumption of the C]O content.36 The physical properties ofthe spent catalysts have been evaluated by XPS and tempera-ture-programmed oxidation (TPO) techniques. The XPS resultsof the oxygen-based species recorded for the different samples,as-received FLG–GO@NDs, pre-treated at 550 �C under He andthe post-reaction FLG–GO@ND composite are summarized inFig. S10.† It shows that the O–H bond of the FLG–GO@NDsremained unchanged, for the sample before and aer thereaction, at about 7.9 at.%, which indicates that hydrogenationof the catalyst, during the course of the reaction, was unlikely tooccur. In addition, the TPO mass spectrum of the post-reactioncomposite displayed no peaks corresponding to H2O during the

J. Mater. Chem. A, 2014, 2, 11349–11357 | 11355

Journal of Materials Chemistry A Paper

Publ

ishe

d on

13

June

201

4. D

ownl

oade

d by

Mon

ash

Uni

vers

ity o

n 25

/10/

2014

08:

48:5

4.

View Article Online

combustion, which suggested that hydrogen generated from thereaction was directly released out from the catalyst.

Conclusion

We have demonstrated in this report an easy and scalablemethod to produce a “sandwich structure” between GO andFLG, where GO plays the role of a dispersant allowing theexfoliation of expanded graphite into a thin FLG structure andprevents the re-stacking of the FLG. The GO layer also providesanchorage sites for adsorbing ND particles through a chargecompensation leading to the high dispersion of this layer on thehybrid support surface. The NDs–GO–FLG composite wasfurther used as a metal-free catalyst in the steam-free dehy-drogenation of ethyl-benzene to styrene. The hybrid compositeexhibits a high and relatively stable DH performance comparedto the other carbon-based and doped iron industrial catalysts.The DH activity per weight unit of NDs was about two timeshigher than the one obtained on the unsupported, which is dueto the high dispersion of the NDs on the GO–FLG surface. TheNDs–GO–FLG also exhibits a relatively high DH stability as afunction of time on stream, up to 150 h. The deactivated catalystcan be efficiently regenerated by air calcination at 400 �C whichallows the recovery of the C]O functional groups and theremoval of the carbonaceous residues deposited during thecourse of the reaction. The results obtained in the present workcan be further extended to other liquid-phase catalytic appli-cations where a highly reactive surface of the GO is required foradsorption of metal nanoparticles.

Acknowledgements

Tung T. T and Lai T. P. would like to thank the European Unionfor the grant through the Freecats project (NMP-2011-22.4). TheSEM and TEM analyses were performed at the facilities of theIPCMS (UMR 7504 CNRS-University of Strasbourg) and T.Romero (ICPEES) is gratefully acknowledged for performing theSEM analysis; P. Bernhardt (ICPEES) is gratefully acknowledgedfor performing the XPS analysis.

References

1 K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang,M. I. Katsnelson, I. V. Grigorieva, S. V. Dubonos andA. A. Firsov, Two-dimensional gas of massless Diracfermions in graphene, Nature, 2005, 438, 197–200.

2 A. K. Geim and K. S. Novoselov, The rise of graphene, Nat.Mater., 2007, 6, 183–191.

3 F. Schedin, A. K. Geim, S. V. Morozov, E. W. Hill, P. Blake,M. I. Katsnelson and K. S. Novoselov, Detection ofindividual gas molecules adsorbed on graphene, Nat.Mater., 2007, 6, 652–655.

4 K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang,S. V. Dubonos, et al., Electric eld effect in atomically thincarbon lms, Science, 2004, 306, 666–669.

11356 | J. Mater. Chem. A, 2014, 2, 11349–11357

5 C. Berger, Z. M. Song, X. B. Li, X. S. Wu, N. Brown, C. Naud,et al., Electronic connement and coherence in patternedepitaxial graphene, Science, 2006, 312, 1191–1198.

6 B. Dlubak, M. Martin, C. Deranlot, B. Servet, S. Xavier andR. Mattana, Highly efficient spin transport in epitaxialgraphene on SiC, Nat. Phys., 2012, 8, 557–561.

7 N. C. Bartelt and K. F. McCarty, Graphene growth on metalsurfaces, MRS Bull., 2012, 37, 1158–1165.

8 S. Park and R. S. Ruoff, Chemical methods for theproduction of graphenes, Nat. Nanotechnol., 2009, l4, 217–224.

9 W. Gao, L. B. Alemary, L. Gi and P. M. Ajayan, New insightsinto the structure and reduction of graphite oxide, Nat.Chem., 2009, 1, 403–408.

10 S. Stankovich, D. A. Dikin, R. D. Piner, K. A. Kohlhaas,A. Kleinhammes and Y. Jia, Synthesis of graphene-basednanosheets via chemical reduction of exfoliated graphiteoxide, Carbon, 2007, 45, 1558–1565.

11 J. T. Robinson, F. K. Perkins, E. S. Snow, Z. Wei andP. E. Sheehan, Reduced graphene oxide molecular sensors,Nano Lett., 2008, 8, 3137–3140.

12 I. Janowska, F. Vigneron, D. Begin, O. Ersen, P. Bernhardt,T. Romero, et al., Mechanical thinning to make few-layergraphene from pencil lead, Carbon, 2012, 50, 3106–3110.

13 D. M. Kosynkin, A. L. Higginbotham, A. Sinitskii,J. R. Lomeda, A. Dimiev, B. K. Price, et al., Longitudinalunzipping of carbon nanotubes to form graphenenanoribbons, Nature, 2009, 458, 872–876.

14 L. Jiao, L. Zhang, X. Wang, G. Diankov and H. Dai, Narrowgraphene nanoribbons from carbon nanotubes, Nature,2009, 458, 877–880.

15 L. Jiao, X. Wang, G. Diankov, H. Wang and H. Dai, Facilesynthesis of high-quality graphene nanoribbons, Nat.Nanotechnol., 2010, 5, 321–325.

16 I. Janowska, O. Ersen, T. Jacob, Ph. Vennegues, D. Begin,et al., Catalytic unzipping of carbon nanotubes to few-layergrapheme sheets under microwaves irradiation, Appl.Catal. A: Gen, 2009, 371, 22–30.

17 L. Jaber-Ansari and M. C. Hersam, Solution-processedgraphene materials and composites, MRS Bull., 2012, 37,1167–1175.

18 C. Shih, A. Vijayaraghavan, R. Krishnan, R. Sharma,J. H. Han and M. H. Ham, Bi- and trilayer graphenesolutions, Nat. Nanotechnol., 2011, 6, 439–445.

19 Y. Hernandes, V. Nicolosi, M. Lotya, F. M. Blighe, Z. Sun,et al., High-yield production of graphene by liquid-phaseexfoliation of graphite, Nat. Nanotechnol., 2008, 3, 563–568.

20 J. N. Coleman, Liquid exfoliation of defect-free graphene,Acc. Chem. Res., 2013, 46, 14–22.

21 A. O'Neill, U. Khan, P. N. Nirmalraj, J. Boland andJ. N. Coleman, Graphene dispersion and exfoliation in lowboiling point solvents, J. Phys. Chem. C, 2011, 115, 5422–5428.

22 U. Khan, A. O'Neill, M. Lotya, S. De and J. N. Coleman, High-concentration solvent exfoliation of graphene, Small, 2010,6(7), 864–871.

This journal is © The Royal Society of Chemistry 2014

Paper Journal of Materials Chemistry A

Publ

ishe

d on

13

June

201

4. D

ownl

oade

d by

Mon

ash

Uni

vers

ity o

n 25

/10/

2014

08:

48:5

4.

View Article Online

23 M. Lotya, P. J. King, U. Khan, S. De and J. N. Coleman, High-concentration, surfactant-stabilized graphene dispersions,ACS Nano, 2010, 4, 3155–3162.

24 M. Lotya, Y. Hernandez, P. J. King, R. J. Smith, V. Nicolosi,L. S. Karlsson, et al., Liquid phase production of grapheneby exfoliation of graphite in surfactant/water solution, J.Am. Chem. Soc., 2009, 131, 3611–3620.

25 A. A. Green and M. C. Hersam, Solution phase production ofgraphene with controlled thickness via densitydifferentiation, Nano Lett., 2009, 9, 4031–4036.

26 D. Li, M. B. Muller, S. Gilje, R. B. Kaner and G. Wallace,Processable aqueous dispersions of graphene nanosheets,Nat. Nanotechnol., 2008, 3, 101–103.

27 J. Kim, L. J. Cote, F. Kim, W. Yuan, K. R. Shull and J. Huang,Graphene oxide sheets at interfaces, J. Am. Chem. Soc., 2010,132, 8180–8186.

28 L. J. Cote, J. Kim, V. C. Tung, J. Luo, F. Kim and J. Huang,Graphene oxide as surfactant sheets, Pure Appl. Chem.,2011, 83(1), 95–110.

29 S. H. Lee, D. H. Lee, W. J. Lee and S. O. Kim, Tailoredassembly of carbon nanotubes and Graphene, Adv. Funct.Mater., 2012, 21, 1338–1354.

30 H. Hu, Z. Zhao, Q. Zhou, Y. Gogotsi and J. Qiu, The role ofmicrowave absorption on the formation of graphene oxide,Carbon, 2012, 50, 3267–3273.

31 V. N. Mochalin, O. Shenderova, D. Ho and Y. Gogotsi, Theproperties and applications of nanodiamonds, Nat.Nanotechnol., 2012, 7, 11–23.

32 Y. Sun, Q. Wu, Y. Xu, H. Bai, C. Li and G. Shi, Highlyconductive and exible mesoporous graphitic lmsprepared by graphitizing the composites of graphene oxideand nanodiamond, J. Mater. Chem., 2011, 21, 7154–7160.

This journal is © The Royal Society of Chemistry 2014

33 D. M. Jang, J. Myung, H. S. Im, Y. S. Seo, Y. J. Cho, C. W. Lee,J. Park, A. Y. Jee and M. Lee, Nanodiamonds asphotocatalysts for reduction of water and graphene oxide,Chem. Commun., 2012, 48, 696–698.

34 D. M. Jang, H. S. Im, J. Myung, Y. S. Seo, Y. J. Cho, H. S. Kim,S. H. Back, J. H. Park, E. H. Cha and M. Lee, Hydrogen andcarbon monoxide generation from laser-inducedgraphitized nanodiamonds in water, Phys. Chem. Chem.Phys., 2013, 15, 7155–7160.

35 J. Zang, Y. Wang, L. Bian, J. Zhang, F. Meng, Y. Zhao, R. Lu,X. Qu and S. Ren, Graphene growth on nanodiamonds as asupport for a Pt electrocatalyst in methanol electro-oxidation, Carbon, 2012, 50, 3032–3038.

36 J. Zhang, D. S. Su, R. Blume, R. Schlogl, R. Wang, X. Yang andA. Gajovic, Surface chemistry and catalytic reactivity of ananodiamond in the steam-free dehydrogenation ofethylbenzene, Angew. Chem., Int. Ed., 2010, 49, 8640–8644.

37 Q. Wang, N. Plylahan, M. V. Shelke, R. R. Devarapalli, M. Li,et al., Nanodiamond particles/reduced graphene oxidecomposites as efficient supercapacitor electrodes, Carbon,2014, 68, 175–184.

38 A. C. Ferrari and D. M. Basko, Raman spectroscopy as aversatile tool for studying the properties of graphene, Nat.Nanotechnol., 2013, 8, 235–243.

39 A. C. Ferrari, J. C. Meyer, V. Scardaci, C. Casiraghi,M. Lazzeri, F. Mauri, S. Piscanec, D. Jiang, K. S. Novoselov,S. Roth and A. K. Geim, Raman spectrum of graphene andgraphene layers, Phys. Rev. Lett., 2006, 97, 187401.

40 Y. Liu, Z. Gu, J. L. Margrave and V. N. Khabashesku,Functionalization of nanoscale diamond powder: uoro-,alkyl-, amino-, and amino acid-nanodiamond derivatives,Chem. Mater., 2004, 16, 3924–3930.

J. Mater. Chem. A, 2014, 2, 11349–11357 | 11357